bpp::OneDimensionOptimizationTools Class Reference

Tools of one parameter-functions optimizations. More...

#include <Bpp/Numeric/Function/OneDimensionOptimizationTools.h>

Public Member Functions
	OneDimensionOptimizationTools ()
virtual	~OneDimensionOptimizationTools ()

Static Public Member Functions
static Bracket	bracketMinimum (double a, double b, Function *function, ParameterList parameters)
	Bracket a minimum. More...
static Bracket	inwardBracketMinimum (double a, double b, Function *function, ParameterList parameters, uint intervalsNum=10)
	Bracket a minimum by a search within thw parameter's bounds. More...
static unsigned int	lineMinimization (DirectionFunction &f1dim, ParameterList &parameters, std::vector< double > &xi, double tolerance, OutputStream *profiler=0, OutputStream *messenger=0, unsigned int verbose=2)
static unsigned int	lineSearch (DirectionFunction &f1dim, ParameterList &parameters, std::vector< double > &xi, std::vector< double > &gradient, OutputStream *profiler=0, OutputStream *messenger=0, unsigned int verbose=2)
	Search a 'sufficiently low' value for a function in a given direction. More...

Static Public Attributes
static double	GLIMIT = 100.0
	Maximum magnification allowed for a parabolic-fit step. More...

Detailed Description

Tools of one parameter-functions optimizations.

For now, contains only one method to bracket a minimum.

Definition at line 93 of file OneDimensionOptimizationTools.h.

Constructor & Destructor Documentation

OneDimensionOptimizationTools()

bpp::OneDimensionOptimizationTools::OneDimensionOptimizationTools ( )

inline

Definition at line 96 of file OneDimensionOptimizationTools.h.

~OneDimensionOptimizationTools()

virtual bpp::OneDimensionOptimizationTools::~OneDimensionOptimizationTools ( )

inlinevirtual

Definition at line 97 of file OneDimensionOptimizationTools.h.

Member Function Documentation

bracketMinimum()

Bracket OneDimensionOptimizationTools::bracketMinimum ( double  a, double  b, Function *  function, ParameterList  parameters  )

static

Bracket a minimum.

Given a function func, and given distinct initial points x1 and x2, this routine searches in the downhill direction (defined by the function as evaluated at the initial points) and returns a Bracket object with new points a.x, b.x and c.x that bracket a minimum of the function. Also returned are the function values at the three points, a.f, b.f and c.f.

Parameters

a,b	Two initial values for the parameter.
function	The function to bracket.
parameters	The parameter to use as a variable.

Returns

A bracket object.

Definition at line 65 of file OneDimensionOptimizationTools.cpp.

References bpp::Bracket::a, bpp::NumTools::abs(), bpp::Bracket::b, bpp::Bracket::c, bpp::BracketPoint::f, bpp::NumConstants::GOLDEN_RATIO_PHI(), bpp::NumTools::max(), bpp::Bracket::setA(), bpp::Bracket::setB(), bpp::Bracket::setC(), bpp::NumTools::sign(), bpp::NumConstants::VERY_TINY(), and bpp::BracketPoint::x.

Referenced by bpp::BrentOneDimension::doInit(), and bpp::GoldenSectionSearch::doInit().

inwardBracketMinimum()

Bracket OneDimensionOptimizationTools::inwardBracketMinimum ( double  a, double  b, Function *  function, ParameterList  parameters, uint  intervalsNum = 10  )

static

Bracket a minimum by a search within thw parameter's bounds.

This implementation follows numerical recipes, page 352, inwards bracketing

Parameters

a,b	Two initial values for the parameter.
function	The function to bracket.
parameters	The parameter to use as a variable.
intervalsNum	number of segments in which [a;b] is splitted.

Returns

A bracket object.

Definition at line 163 of file OneDimensionOptimizationTools.cpp.

References bpp::Bracket::a, bpp::Bracket::b, bpp::Bracket::c, bpp::BracketPoint::f, and bpp::BracketPoint::x.

Referenced by bpp::BrentOneDimension::doInit().

lineMinimization()

unsigned int OneDimensionOptimizationTools::lineMinimization ( DirectionFunction &  f1dim, ParameterList &  parameters, std::vector< double > &  xi, double  tolerance, OutputStream *  profiler = 0, OutputStream *  messenger = 0, unsigned int  verbose = 2  )

static

Definition at line 210 of file OneDimensionOptimizationTools.cpp.

References bpp::ParameterList::addParameter(), bpp::AutoParameter::CONSTRAINTS_AUTO, bpp::AutoParameter::CONSTRAINTS_KEEP, bpp::AbstractOptimizer::getNumberOfEvaluations(), bpp::DirectionFunction::getParameters(), bpp::AbstractOptimizer::getStopCondition(), bpp::DirectionFunction::init(), bpp::AbstractOptimizer::init(), bpp::BrentOneDimension::optimize(), bpp::AbstractOptimizer::setConstraintPolicy(), bpp::DirectionFunction::setConstraintPolicy(), bpp::BrentOneDimension::setInitialInterval(), bpp::DirectionFunction::setMessageHandler(), bpp::AbstractOptimizer::setMessageHandler(), bpp::AbstractOptimizer::setOptimizationProgressCharacter(), bpp::AbstractOptimizer::setProfiler(), bpp::OptimizationStopCondition::setTolerance(), bpp::AbstractOptimizer::setVerbose(), and bpp::ParameterList::size().

Referenced by bpp::ConjugateGradientMultiDimensions::doStep(), and bpp::PowellMultiDimensions::doStep().

lineSearch()

unsigned int OneDimensionOptimizationTools::lineSearch ( DirectionFunction &  f1dim, ParameterList &  parameters, std::vector< double > &  xi, std::vector< double > &  gradient, OutputStream *  profiler = 0, OutputStream *  messenger = 0, unsigned int  verbose = 2  )

static

Search a 'sufficiently low' value for a function in a given direction.

This function performs a similar computation as the lnsrch function defined at page 385 of

NUMERICAL RECIPES IN C: THE ART OF SCIENTIFIC COMPUTING
(ISBN 0-521-43108-5)

without the stpmax argument, since the steps are bounded in another way.

Definition at line 252 of file OneDimensionOptimizationTools.cpp.

References bpp::ParameterList::addParameter(), bpp::AutoParameter::CONSTRAINTS_AUTO, bpp::AutoParameter::CONSTRAINTS_KEEP, bpp::AbstractOptimizer::getNumberOfEvaluations(), bpp::DirectionFunction::getParameters(), bpp::AbstractOptimizer::getStopCondition(), bpp::DirectionFunction::init(), bpp::AbstractOptimizer::init(), bpp::AbstractOptimizer::optimize(), bpp::AbstractOptimizer::setConstraintPolicy(), bpp::DirectionFunction::setConstraintPolicy(), bpp::DirectionFunction::setMessageHandler(), bpp::AbstractOptimizer::setMessageHandler(), bpp::AbstractOptimizer::setOptimizationProgressCharacter(), bpp::AbstractOptimizer::setProfiler(), bpp::OptimizationStopCondition::setTolerance(), bpp::AbstractOptimizer::setVerbose(), and bpp::ParameterList::size().

Referenced by bpp::BfgsMultiDimensions::doStep().

Member Data Documentation

GLIMIT

double OneDimensionOptimizationTools::GLIMIT = 100.0

static

Maximum magnification allowed for a parabolic-fit step.

Definition at line 151 of file OneDimensionOptimizationTools.h.

---

The documentation for this class was generated from the following files:

• Bpp/Numeric/Function/OneDimensionOptimizationTools.h
• Bpp/Numeric/Function/OneDimensionOptimizationTools.cpp